Evaluation of NCEP Global Forecasting System (GFS) Performance in
Predicting Landfalling Tropical Cyclone Rainfall in Korea
Abstract
This paper discusses the performance of the National Centers for
Environmental Prediction (NCEP) Global Forecasting System (GFS) forecast
model in predicting rainfall over the Korean Peninsula caused by
tropical cyclones (TCs). TCs forecasted in GFS were characterized by
overestimated intensity and well-organized inner core structure during
landfall, and the associated TC-induced rainfall forecast errors were
examined from two perspectives: TC tracks and TC-environmental
interactions. The results provide implications in three ways. First, the
predicted TC track is biased to the west, bringing it closer to the
Korean Peninsula landmass and contributing to an overall increase in
overland precipitation. Second, the strongly simulated TC during
landfall predicted rainfall only in a narrow radius around the TC
center. Third, the TCs were characterized by weak upper-level TC outflow
due to the well-developed TC structure, which could increase the
inertial stability of the TC center during jet interaction, leading to
an underestimated precipitation in outer core region of the storm. These
results suggest specific sources of error in the forecast of TC
precipitation over the Korean Peninsula, which can contribute to
improving numerical models and providing forecasters with information
about model characteristics.